LLM
GA
Open source
Gemma 4 26B A4B
Local and on-device deployment under a permissive license
Release2026-04-02
Context262K
Max output—
Parameters26B
Local / VRAMLocal
LicenseOpen source
Overview
Gemma 4 26B A4B is a model in the LLM category from Google DeepMind. It was released in 2026-04-02. The vendor is based in US/UK. Best for: Local and on-device deployment under a permissive license.
Notes: Gemma 4 family released 2026-04-02; first Gemma generation under Apache-2.0.
Hardware: VRAM figures not verified; all weights must fit in memory even if only a few parameters are active.
Strengths
- Fast MoE: only ~3.8B active parameters per token
- Apache-2.0 license
- Same research base as Gemini 3
Limits
- Weaker than the 31B dense model
Specifications
| Vendor | Google DeepMind |
|---|---|
| Category | LLM |
| Release | 2026-04-02 |
| Status | GA |
| License | Open source |
| License name | Apache-2.0 |
| Commercial use | Allowed |
| Parameters | 26B |
| Active parameters | 3.8B |
| Architecture | MoE transformer |
| Context window | 262K |
| Input modalities | Text, Image |
| Output modalities | Text |
| Prompt types | Text |
| Negative prompts | No |
| Max outputs | 1 |
| Local deployment | Yes |
| API access | Yes |
| Pricing | Free weights |
| Confidence | High |
| Vendor country | US/UK |
| Vendor type | Frontier lab |
Pricing & hardware
Free weights
This model can be self-hosted or run locally. API access is available.